Defining Cis And Trans Mapping Windows

Why Mapping Windows Need an Explicit Definition

A molecular QTL analysis tests whether genetic variants are associated with a molecular phenotype, such as expression of a gene. Before testing, the analyst must decide which variant–phenotype pairs are in scope. A search near a gene is commonly called cis; a search farther away, potentially across the genome, is commonly called trans. These labels describe a mapping rule, not a proven biological mechanism.

There is no single universal distance at which a variant changes from cis to trans. A window is an operational choice: it specifies the genomic locations to search for a given molecular phenotype. That choice affects which associations can be discovered, what “cis” means in the reported results, how much computation is required, and how many statistical tests contribute to the search.

How to Specify a Cis Window

A common rule centers the window on a gene’s transcription start site (TSS). Let \(t_g\) be the TSS position for gene \(g\), and let \(W\) be a chosen distance in base pairs. A symmetric TSS-centered window is the interval from \(t_g-W\) to \(t_g+W\). Under an inclusive endpoint convention, a variant at either boundary is included.

Another rule starts from the gene body, whose annotated start and end positions are \(a_g\) and \(b_g\). A gene-body window can extend \(L\) base pairs upstream and \(R\) base pairs downstream, producing an interval from \(a_g-L\) to \(b_g+R\). The upstream and downstream directions refer to transcription, so the genomic coordinates used for those flanks depend on whether the gene is on the forward or reverse strand.

These definitions are not interchangeable. A TSS-centered window is anchored to one point; a gene-body window includes the span of the annotated gene and can be asymmetric. A protocol should state which is used, the chosen distance or flank lengths, and how endpoints are handled. The TSS also depends on the transcript annotation: genes can have multiple transcripts with different start sites. An analysis should identify the annotation and transcript-selection rule rather than silently treating every gene as having one unambiguous TSS.

Coordinates must be tied to a reference genome build. A position such as 10,000,000 is meaningful only in relation to a particular genome assembly and coordinate convention. For example, one may specify that variant positions and gene coordinates use the same build and 1-based reference coordinates. If variants and annotations use different builds, positions must be harmonized before applying the window rule. Otherwise, a variant can be assigned to the wrong interval even when the stated distance rule is clear.

Defining the Trans Search

Once the cis rule is defined, a straightforward pair-specific convention calls variants outside that gene’s cis interval trans for that gene. This includes distant variants on the same chromosome as well as variants on other chromosomes. Some studies instead restrict “trans” to different-chromosome pairs or impose a separate minimum distance. Those conventions can be useful, but they produce different search sets and must be reported explicitly.

Classification is relative to the target gene. A variant can be inside the cis window for one gene and outside the cis window for another. Overlapping windows make this especially clear: the same variant may be tested as cis for one phenotype and trans for a different phenotype. Thus “cis variant” is often shorthand; the more precise statement is “cis for gene \(g\) under the specified window rule.”

Search-Space Size, Computation, and Power

For each molecular phenotype, the number of variant–gene tests depends on the chosen window and on which variants are available. A narrow cis interval usually contains fewer variants than a broad interval, reducing the number of tests and the computational work required. A genome-wide trans search can involve far more pairs: in a simple design, each variant may be tested against many molecular phenotypes. The number of pairs can become large even when the number of donors is unchanged.

Search-space size also affects statistical power when an analysis controls the chance of any false positive across the tests. To see the basic relationship, suppose \(M\) null hypotheses are tested and each test has a valid p-value. If each test uses a threshold \(\alpha/M\), the probability of at least one false positive is at most \(\alpha\). This follows because the probability of the union of false-positive events is no greater than the sum of their probabilities:

This bound does not require tests to be independent, though it can be conservative when tests are correlated. It shows why, under this particular family-wise error rule, a larger search requires a smaller per-test threshold. A smaller threshold makes it harder for a true association to be declared significant, all else equal. The precise power still depends on sample size, effect size, phenotype variability, variant frequency, linkage disequilibrium, and the chosen analysis procedure. A narrow window does not automatically have higher power if it excludes the causal variant or a useful proxy.

Biological Interpretation Depends on the Boundary

A broader cis window can include regulatory elements that are far from a TSS, including elements between genes or in neighboring genes. This can increase the chance of including a variant that is biologically relevant to the target phenotype. It can also make a “cis association” less local in the everyday sense and complicate interpretation when nearby genes share regulatory regions or are in linkage disequilibrium. An associated variant in a broad window does not necessarily act through the nearest gene.

A narrower window offers a more restrictive definition of local association, often with fewer tests and a simpler positional description. But a fixed boundary can exclude a functional variant beyond it. Such a variant may then be labelled trans by the operational rule even if it acts through a nearby regulatory element. Conversely, being within a cis window does not prove a local mechanism. Genomic distance is a useful search convention, not a causal test.

Window choices also affect comparisons across studies. Two studies can use the same word, “cis,” while testing different intervals, reference builds, or transcript annotations. Their reported counts of cis QTLs are not directly comparable unless those choices are aligned or their differences are made explicit. A transparent report gives the build, annotation, TSS or gene-body rule, window size or flank lengths, endpoint convention, and trans definition.

Common Mistakes

• Treating one distance as a biological law. A variant just outside a chosen boundary is not known to be biologically unrelated to the gene. The boundary is an operational convention.
• Calling every trans pair different-chromosome. Some definitions include distant variants on the same chromosome; others do not. State whether trans is the complement of the cis window or uses a separate rule.
• Ignoring strand and transcript annotation. A TSS is transcript-specific, and upstream/downstream flanks depend on strand. Specify the annotation and how the TSS is selected.
• Comparing windows across incompatible coordinates. Variant and gene positions must use the same genome build and coordinate convention before distances are calculated.
• Equating a cis association with direct regulation. Position alone does not establish mechanism, identify the causal variant, or prove that the target gene mediates an effect.